No such file or directory @ rb_sysopen about html-proofer

[doktorbro]

After adding the missing file the issue is gone. But it is just a workaround. The question is, why Proofer tries to proof an non-existent file?

[gjtorikian]

It's more a feature then a bug. Basically, like every browser ever, Proofer expects there to be a top-level index.html file when you navigate to a folder. For pages with clean URLs (or just using Jekyll), checking for an index.html is really handy.

The workaround suggested was to add data-proofer-ignore to the anchor. I could in the future add an option to disable assumption of index.html but I have not seen it as a problem yet.

[parkr]

Basically, like every browser ever, Proofer expects there to be a top-level index.html file when you navigate to a folder.

If the link is to a folder, then it can be any number of things (index.html, index.htm, index.shtml, index.php, index.xml, etc) depending upon the server setup. You're not assuming there's going to be an index.html file for every folder level, right? Just for those directories being linked to?

[gjtorikian]

shifty eyes

html-proofer/lib/html/proofer/checkable.rb

Line 108 in 4904bf8

file = File.join path, "index.html" if File.directory? File.expand_path file, @check.src

You're not assuming there's going to be an index.html file for every folder level, right? Just for those directories being linked to?

Yes. But now your mention of all those other file types has me in a tizzy. 😩

[parkr]

Yes. But now your mention of all those other file types has me in a tizzy.

I think HTMLProofer can assume either html, htm, or xhtml. But honestly... should this be a tool for legacy sites? 😉

shifty eyes

If @check.src is a dir, then I think it's safe to assume you should check for a common index file. Maybe that can be a config option?

[doktorbro]

@gjtorikian The error occurs while you load the source files with outgoing links before you start to check anything: https://github.com/gjtorikian/html-proofer/blob/master/lib/html/proofer.rb#L25 Nevermind.

[nschonni]

Yeah, I just ran into this one as well https://travis-ci.org/nschonni/wet-boew/builds/32812844#L2463
Slight difference was that there is no link to a folder root, but just rally long query string plus anchor link (I think that is what is tripping it, but there is no trace back)

[doktorbro]

@nschonni Can you please provide the exact link?

This is your situation now, right?

# does not exist
dist/demos/texthighlight/index.html

# do exist
dist/demos/texthighlight/texthighlight-fr.html
dist/demos/texthighlight/texthighlight-en.html

[nschonni]

@penibelst I believe the link it's having trouble with is ?txthl=avian%20influenza+world+cook+flu-like%20symptoms+Don't%20Forget...+causes%20sickness%20in%20birds,%20it%20can%20also%20infect%20people.#example from here https://github.com/wet-boew/wet-boew-dist/blob/4b3ea7429223726c674350737cd879bc31992c3e/demos/texthighlight/texthighlight-en.html#L136

You can grab the prebuilt site if you want to test it out https://github.com/wet-boew/wet-boew-dist

Oddly enough, if I add :followlocation => false and :disable_external => true then it runs, although then other missing trailing slashes are found, but it no longer crashes.

[doktorbro]

I think #93 fixes all the mentioned issues.

[doktorbro]

@nschonni The version 1.3.0 fixes your issue.

[nschonni]

Thanks @penibelst! It's not crashing anymore, but it just stalls out completely https://travis-ci.org/nschonni/wet-boew/builds/33367787
I'll see if I can find a combo of options that lets it run again.

[doktorbro]

@nschonni Can you please start Proofer with only one file?

HTML::Proofer.new("example.html").run

[nschonni]

Yeah, running a smaller set help, but it's taking almost 10minutes to run on less than 200 pages now. Getting false positives for the href links, not sure if it's tripping on HTML5 elements, but this is the wrong thread for that 😉

[doktorbro]

@nschonni External URLs are unpredictable. Try the timeout option:

HTML::Proofer.new(dir, {
  :timeout => 10 # seconds
}).run

If you see false positives, please open a new issue with examples.